Welcome to the sixty-third episode of Access On, the National Federation of the Blind's Technology podcast.
Episode
Listen to the sixty-third episode of the Access On podcast (Browser).
Or listen on your preferred podcast platform.
Timestamps
Here's what you can hear and when:
- Register for our Access On webinar on 1Password 0:00
- Overview of AI in employment 1:53
- Working with Google AI tools in the workplace 8:00
- Microsoft tools in the workplace 29:04
- Microsoft PowerPoint and AI 51:49
- Using AI to take notes 1:06:56
- Closing and contact info 1:14:23
Transcript
Speaker 1:
Live life you want.
Speaker 2:
Access on.
Jonathan Mosen:
Welcome to Access On, the technology podcast of the National Federation of the Blind. This week, time's running out to register for our February 24 Access On 1Password webinar and we bring you highlights about webinar covering using AI tools to be more productive on the job.
Thank you for being back with us for episode 63 of Access On. It's Jonathan Mosen at the Jernigan Institute in Baltimore, Maryland. Before we get onto the main part of the podcast, which is looking at this ever-changing, increasingly capable world of AI in employment, I want to remind you of our Access On webinar, which is fast approaching as we publish this podcast.
It's going to be on Tuesday, February the 24th between 2:00 and 3:30 PM Eastern. This is by listener request on 1Password, the password management tool that also allows you to store other sensitive data and retrieve it and easily insert it wherever it needs to be inserted.
We'll also cover how to use 1Password to perform two-factor authentication to make your accounts more secure and also this emerging much safer world of giving up on passwords all together and where possible going to pass keys instead. It's all going to be covered in our Access On webinar. If you can't make it but you register, then you will get a recording of the full event. But if you do turn up, that's great because it means that you can ask questions in the Q&A part and perhaps share your own experiences with 1Password if you're a user already.
To register, head on over to nfb.org/cena, that's nfb.org/cena and register there and we look forward to seeing you on February the 24th.
Late last year, we held a webinar on a hot topic and this is using AI tools on the job to get things done more productively. And of course there are also some accessibility benefits of using AI where there may be content that's not as accessible as it ought to be or you've got some visual content that you would like described. So this webinar covers some of these points and I hope that you find the extracts that we're playing from it of interest.
Let me start with a striking fact for you. Three quarters of us who are knowledge workers are already using AI at work today with nearly half beginning just in the past six months. Research from the Federal Reserve Bank of St. Louis found that workers using generative AI saved 5.4% of their work hours in a typical working week. The impact varies dramatically by profession.
Nielsen Norman Group studies found that programmers using AI completed more than double the projects per week. Business professionals wrote 59% more documents per hour. 90% of AI users said that it helped them save time, and 85% say that it helps them focus on their most important work. 84% feel that it allows them to be more creative and perhaps most importantly, 83% says that AI has made their work more fun.
As the mass adoption by blind people of tools like Be My AI and hardware like the Meta glasses have demonstrated, AI has the potential to improve blind people's access to information. AI can now, for example, generate descriptions of charts and graphs or images using reports, training or presentations. When a page contains images that don't have descriptions, we know what used to happen, right?
Screen readers would traditionally just say image or unlabeled graphic, which was helpful not. Now, the automatic image description features that are found in several browsers can create descriptions for millions of images using machine learning to identify text, objects and context.
AI is democratizing the creation of accessible content itself. New tools enable blind users to create our own custom data representations, but at the National Federation of the Blind's Center of Excellence in Nonvisual Accessibility, it's our job to call it as we see it so you're fully informed and as anybody who uses AI knows, it has significant limitations.
AI systems generate incorrect information up to 20% of the time, and the thing is they present these errors with complete confidence. This isn't just about minor mistakes as well. We have seen attorneys sanctioned for submitting legal briefs with completely fictional cases. They were generated by the AI and got the lawyers into some strife.
Second, there is the skills crisis. Only 6% of employees feel very comfortable using AI in their roles, and this disconnect between rapid adoption and actual preparedness creates real risk, but it also does create real opportunities for blind people who choose to invest the time in becoming AI savvy. We know that as blind people, we are too often limited by low expectations. If we're job hunting and we can demonstrate proficiency with AI, that's a skill that's in high demand right now and it could cause an employer to take another look at our applications.
Third, there are substantial security and privacy concerns. A staggering 84% of workers using generative AI admitted they've publicly exposed their company's data in the last three months. And finally, unlike humans, AI can't adapt to evolving situations with true creativity. It can't make ethical judgements that balance business goals with societal impact, and it lacks the emotional intelligence needed for nuanced human interactions. So how do we harness the benefits while minimizing the risks?
Well, first, explicitly give AI permission to say, "I don't know." This simple technique can drastically reduce false information. Never pressure an AI system to provide an answer when it indicates uncertainty. Second, as we've highlighted in previous AI webinars that we've conducted, use specific, detailed prompts.
Provide clear, detailed instructions that guide the AI with specific context, desired details, and cited sources. They help prevent the system from making unwarranted assumptions or fabrications. Third, always verify critical information. Trust but verify. For important tasks, run the same prompt multiple times and compare the output because inconsistencies can often indicate potential hallucination. Fourth, maintain human review as an essential safeguard. Human fact-checkers can identify and correct inaccuracies that AI may not recognize.
So those are the four key points. We are all learning together, but it seems clear that the path forward isn't about choosing between human workers and AI. It's about thoughtful integration. To get into the practicalities, let's first hear from Stephen Polacek from the Maryland Department of Disabilities who's going to be talking about some of the work they're doing with Google AI tools.
Stephen Polacek:
Thank you, Jonathan. So a couple things I want to state before I get into exactly what we're doing. As Jonathan said, AI is still in that experimental phase. We're still trying to figure out what its capabilities are, how to judge its accuracy, how to improve its accuracy. So everything I'm talking about today is experimental. Nothing has gotten to a point where we can confidently release it as a tool, but these are indications of what we're looking at and trying to do with AI and the state. So I can't speak for all of the projects or even some of them, especially for other agencies because all of them are doing a lot with it.
Most are using tools like Notebook that I'll talk about shortly to try and improve their casework or their accuracy in cross-referencing cases. They're trying to use AI in terms of the usual methods you hear about with writing, changing your writing style, improving plain language, that kind of thing, but ultimately, the goal of the state and the goal of the AI office and the Department of IT is to support other agencies in the adoption and guide the safe, ethical and secure use of it.
So everything I'm going to talk about is using Gemini, which is Google's AI because the state is a Google Workspace, so we're using ... most of us are using Gmail and Google Docs and such for our work. So since Gemini's built into all of that, that's why we're using that. Some other agencies are using the other tools that you'll hear about today, but I'm mostly familiar with Gemini, so I'm going to speak about that and leave Copilot and such up to my colleagues here.
So one of the main things that we have been working with is NotebookLM, which is a subset of Gemini. And the nice thing about it is that it was originally intended as a study aid, but it allows you to add sources to it and control its access that it only pulls from those sources. It still thinks the same way Gemini does or builds its logical conclusions that way, but it will only use the sources you provide it and if those sources don't have an answer, it will tell you.
So it's a way to enhance the accuracy of what you're trying to pull from it and also use it to collate a lot of different resources into one area. In our case, our original intent with it was to try and use it in order to help with our testing. When we test websites and other software and what have you, instead of having 12 different tabs open trying to cobble a solution together or a recommendation for a solution, I could have those into the Notebook and then have Notebook spit out a working example or a fixed example.
We've since learned that it's not really capable of doing that. Notebook in and of itself isn't, and like Gemini as a whole, isn't very good at remediation and this is the problem with AI and accessibility right now. As we also saw at a recent conference, I'm enabling, it isn't at the point where it can go and fix code. It doesn't understand context still.
So it'll pull from different examples and put together something that might in theory work, but is usually missing either particular variables or particular parts of the HTML or CSS or JavaScript and just won't function. It requires editing, so it requires that human verification to make it work.
Not so useful when we're trying to explain how to fix something to another agency or their vendor. However, what we have been finding it useful for now is explaining concepts to people or explaining what an issue is and where to find the resources to fix it. It's acting as more as an advisor because since I and my colleague, Dan, are the ones who have added the sources to it, we know those sources are correct and we've marked them as to different types.
We don't have a way in Notebook yet to filter to say pull from only these, but you can tell it, "I only want to check these sources. So any ones I've marked with documents can be used for answering questions about document accessibility. Ones I've marked for website testing, same idea. Ones that are just about standards. Again, same idea.
If the question is just about standards or if the user unchecks the sources that are not marked with that, they can use those to get the answer to their question rather than trying to try and find it through Google or doing another type of internet web search or trying to find an answer in either help article or from another technical space where they might not know how to navigate or understand what the person's talking about because they're talking specifically about code instances. This way we give our users a potential way to get information about how to make something accessible that is pulling from resources we know are correct because they're ones we use and we've verified as correct.
Reduces the chances for hallucination. It gives our users a way to really understand what they're looking for in a more plain language function. The reason for that is because you can also customize how Notebook responds. In our case, so there's a default, then there's a learning guide and then there's a custom. That custom lets you use 500 characters to define how it works. So in our case, we've told that respond more as a technical advisor, provide concise solutions and reasons and use plain language when explaining what the issue is and how to fix it.
We can also change the response length to default or longer or shorter. We've chosen shorter because longer and default tend to spit out multiple pages and cite every possible source, which got overwhelming, but now we have a way that we can give short responses with citation to where the correct technique is and plain language version of that. We're currently trialing out with an intern or two interns at our department to see how they find it functioning and what further refinements we can do with it.
But we're hoping in the future that we can open that up to others across the state to use as a way to deal with accessibility issues and answer their questions easily before having to either come to us or before trying to search through the web without the background knowledge to filter out what might work and what might not for their situation.
So Notebook is the main one we're using in that regard. It has other functionalities. Again, I mentioned it's more of a study guide, so there's ways to make overviews of your sources or reports or you can even develop flashcards, quizzes and mind maps to test yourself with. But other agencies I know have been looking into it as a way to collate information about cases.
I can't talk much about that because that's another department, but they have been trying to use it as a way to teach new hires that come in about how to handle a case or how to handle a specific portion or a specific issue that might come up or how cases generally flow from stage 1 to stage 10. So they've been using it as a training tool for new hires and that's a way AI is being used in employment in that capacity.
The other area that we've been looking at is Gems. So Gemini has these smaller instances called Gems where it's still using Gemini, the app, as its base. So any question or any other information you put into it is going into Gemini, so you do need to be careful with what you're allowing it access to and where you're using it.
However, the nice thing about it is that it's again something you can kind of tailor. Unlike Notebook, which takes 300 sources, you can only add 10 to a Gem, but Gems you can add actual instructions to. So you can tell it, I want you to act this way. Behave this way. I want you to look at this file or when you're prompted to look at a file, follow through this checklist and give back these answers.
I first learned of this from the AI office who found an article from Colorado and a state employee there who had developed a Gem that they are made into an accessibility checker and we were looking into that and saw how they set it up. So we're looking into ways that we can build additional ones for other types of documents because that one was just a more generic one and we're looking at setting up one for Google Docs, Google Sheets, Google Slides, so that you have an accessibility checker because the Gems can be used while you're inside a Google Doc or inside Google Drive or whatever.
And those Gems, if we set them up correctly, give them the right instructions, look at this file, check for alt text, check your color contrast, check for headings, an Excel sheet or sorry, Google sheet. Check for proper table headings that things are formatted as tables. In the case of a slide deck, check that speaker notes are there, or remind the user that any speaker notes should ... remind the speaker to give more descriptive language for what's onscreen so that things are explained to their audience correctly. And so by building these checkers, we're now giving people in the state away to have an accessibility checker and a tool that doesn't have one built in.
You might have heard of Grackle, which is used in Google Workspace as an accessibility checker, but that's an external plugin and that isn't always looked upon nicely in the state, so it's not in the high use. But by building a Gem that could maybe do this kind of thing, it's possible that we could give everyone the state away to use an accessibility checker with something that doesn't have one. And so rather than exporting things or having to create multiple versions or whatever, we're able to give them a way to do accessibility work and do testing and remind them about specific challenges, especially for those working in Teams with those with disabilities that A, you need to do X, Y, Z quick.
That is very experimental right now. We've only just started building it out and we're having some challenges with refining it and getting it to answer consistently because, as I said, it is using Gemini as a whole as its base. So sometimes it's pulling from not our sources and that has been proving a bit of a challenge, especially also because you have to point it directly at a file. Sometimes it uses the files as reference and checks those and gives us an answer about those. So still taking some work, but AI is letting us build these kinds of tools that we didn't have before.
The other thing with Gems that we're trying to do is help us build a tool that will help us fill out our reports faster. So in our case, when we do our testing, we write a report giving the issue we found, how to fix it, which guideline it's violating or which one they should look at to find the techniques to fix it, and then its priority level.
Now with this tool and some scripting, we've added to it because we can tell it to run a script that we've written. We can put in the issue and we've given it a bunch of our old reports from another sheet to look at and pull from. It will use those as examples in order to find the correct solution and priority level and guideline and fill in our report for us.
So we're using it as a way to quickly reference old reports and make sure that we're also consistent in how we're reporting how to fix things. Because sometimes with accessibility testing, that can be a little bit of a challenge either because it's not quite the same issue that we found before, but in other times, it's because we thought this might be a better solution or we've learned new things since that last report, which was maybe last year or two years ago. Who knows?
So Gems give us a way to again reference older material in a way that we couldn't really do automatically before and also is helping speed up our work because we're able to pull a report together faster as two people just giving a list of issues rather than writing out the entire report.
Finally, Google AI Studio, which is their vibe coding tool as for lack of another way to put it, but you can use AI Studio, you can give it prompts, tell it, "I want to build an app to do this. It should have these functions in it. I want the UI to have this. Oh, we need to correct this because there's an error here." That kind of thing. And this is where we've been exploring how we can make traditionally inaccessible things accessible.
So one of my personal projects has been looking at can we make something that can look at data visualizations or look at a chart or graph and let the user either generate summaries or pick set questions or answers and set summaries from it, and create dynamic interactions with it in order to make that more accessible because previously, with data visualization and any large data set, if you look at any of the data dashboards that are coming out across the state, those are very hard to make accessible in a meaningful manner.
We have that data in a table, we then put it into a chart. We try and give people a quick summary or a way to interact with it and get just a quick glimpse of what it means. But then in order to be compliant, we say put it back into the chart or back into the table, sorry, and that table, it might have filters, it might have cell heading or row headings or column headings, but it still requires users to move through all of that data themselves, parse it, and then draw those correlations and those relationships together by themselves.
That puts a lot of effort on the user. If we could create a tool that could take that data, use that chart or that table, pull from it, generate maybe several types of summaries like here's trends in this specific year, or here's how month of last year compared to month of this year, any of those kinds of things or maybe even potentially on the fly alterations or something, any kind of way that we can make users get that summary or get those quick understanding of the data set in a way they couldn't will go a long way towards making all these new data dashboards and data visualizations more accessible so that a user who's using a screen reader or has low vision is using a magnifier.
Instead of having to parse through this 50, 100, 200 however size chart and columns or table, they can go to this tool say, I want a summary of this part of the data or compare this year to last year and get that answer quickly in the same way a sighted person can look at a chart and make those two comparison points in an instant.
It is completely ... Don't get too excited for any of you who are into data. It's not there yet. Like I said, this is more proof of concept that I'm building to try and go to the AI office or the office of enterprise data and say this is something that we can build and make. I've just started talking with the governor's innovations team about the same project because they've been doing some things with data visualization to make more interactive and meaningful dashboards and seeing how we can incorporate the ideas to make sure that all this new information that's coming out from the state is more accessible and more usable by everyone.
So that's what my team is doing with AI. In terms of the larger department, there's a lot of different things. Most people, again using it the same ways as you hear about everywhere else when you hear about AI, helping with polishing emails or polishing reports or changing report or a document into different format like a standard operating procedure, things like that or pulling from information, but there are some other trials that are being looked at.
One of the main concerns that the department is looking to try and address is how AI is used in employment and in hiring because there have been scholarly articles and other articles about the bias AI has against disabilities in hiring and how we can combat that issue.
So we've been reading those and looking into how AI might be used in the state in terms of hiring and how we can make sure to mitigate potential issues with it misassigning behaviors for people that it's interacting or reading the resumes of or if it's used during the hiring process during interviews that we're avoiding the potential for AI to misinterpret how a person with a disability is acting on camera or has put certain language in their resume that the AI might filter out because it doesn't match that to the criteria it was told.
So that's one of our main concerns and one of the things that others in the office are really working on how to mitigate or even alter AI to include these things as potential pluses instead. So I don't have much information about that because again, it's a lot of experimentation at that point of how can we make these things work, but it is an area that AI has seen a lot of use in and that we are aware of and trying to work on. So that's everything that we're doing with AI, but like I said, the rest of the state, there's a huge push to use it in every agency. So if you're at all curious about how maybe Department of Aging or Department of Health or Department of Transportation is using it, I'm sure you can find someone there to discuss it.
The state as a whole has put in a community of practice that meets every month. I presented at once, been listening to it for the last few months as well. There is a couple different interagency chats where we're all sharing ideas and resources and what we're finding has worked and what doesn't work and warning each other about, hey, you need to make this accessible. That's mostly me and my team, but at least someone is. So we're making sure that the word gets out there about how to use AI and make sure that you're not just throwing stuff at the wall to see what sticks. You're also trying to consider the end result and how the end result can be made more accessible.
Jonathan Mosen:
Stephen Polacek from the Maryland Department of Disabilities talking about some of the work that they are doing in the Google ecosystem with respect to AI in the workplace. There's more to come as we take a look at this topic. We'll turn to Microsoft next when Access On continues.
Speaker 5:
Do you enjoy our podcast? Do you benefit from the advocacy we do every day? Please help us continue to provide this kind of programming and more. We need your help. Visit nfb.org/donate or call 4106-599-314 extension 2430.
Speaker 6:
Advocacy doesn't end with awareness. It ends with access. Under ADA Title II, public entities must ensure their digital content is accessible by April 2026. That includes existing documents people rely on every day. Many agencies underestimate the scope, thousands of inaccessible documents, limited time, limited budgets. The good news is that large scale remediation is possible. For clear guidance on what the rule actually requires, visit www.title2.info.
This message is brought to you by Numa Solutions. We've remediated hundreds of thousands of pages in days aligned with WCAG 2 AA guidelines at enterprise scale and speed with without sacrificing usability for blind users. Accessibility isn't a privilege, it's a right,
Jonathan Mosen:
And this is Access On, bringing you highlights of our webinar on AI and employment. Just want to add that yes, at the National Federation of the Blind, we share the concern around the impact of AI algorithms on employment and choosing people for the job. We do work with all of the large language model manufacturers on their misconceptions about disability and we feel we've made a lot of progress. But there are those smaller language models that may be a little bit rogue and we keep seeking evidence of what's happening in that space. And if you have any evidence, if you think that AI algorithms are disadvantaging you on the job hunt, we would love to hear further about that.
So chances are that you're either working in a Google shop or a Microsoft shop, and that is why we've focused on Google and Microsoft offerings in this employment-related webinar. We will come back another time and talk about the exciting things that OpenAI and Claude are doing. But even if you have Microsoft 365 at your workplace, it doesn't automatically follow that you have Copilot. Copilot costs your organization extra and Microsoft charges a license per user. If Copilot isn't rolled out generally, depending on what you do, you may be able to make a case to the organization you work for as a blind person that it will improve your output.
To fully leverage Copilot's capabilities, open or save your documents to OneDrive or SharePoint online instead of a local drive. Copilot can only analyze content it can access and in the desktop apps, that means files need to be in the cloud. Assuming you have a license, Microsoft 365 Copilot is integrated across many office applications that you already use, including Word, Excel, Outlook and Teams. PowerPoint integration exists as well, and Kennedy will discuss that after I finished.
Copilot appears as a sidebar or a pane within these apps. You can open the Copilot pane at any time using a keyboard shortcut, for example, in Word or Excel, you can press Alt H, then F, and then X in sequence. The focus will move to an input area where you can type a question or a command for Copilot. You can use F6 or Shift + F6 to cycle focus between the document content and the Copilot pane, and that's useful for moving back and forth between reading your document and entering prompts.
Microsoft also provides a standalone Copilot web application and that's available by visiting copilot.microsoft.com in a web browser and signing in with your Microsoft 365 work account. It gets a bit confusing because in Microsoft land, everything's called Copilot, so we're talking specifically about the Copilot for office in a work situation.
The Copilot web app, which is sometimes called Copilot Chat or the Microsoft 365 Copilot app, offers a central place to interact with the AI across all of your data, even if you're away from the office and away from your desktop apps. For example, from a browser, you could ask Copilot to summarize a document or find information across your emails and SharePoint files all without opening each individual app, hence the need to save to the cloud when you want to work with Copilot.
The web interface is consistent with the in-app Copilot and it supports full navigation and screen reader compatibility. This flexibility means you can use Copilot in the way that works for you at any given time, and you can be inside Word or Excel or Outlook when you're editing your documents or you can be in the browser or in a mobile app when you need to be working in that environment. So you can even use this when you don't have Office installed on a device.
What I'd like to do is walk you through several real world scenarios for using Copilot on the job as a blind professional, and let's start with a common challenge making sense of visual charts or graphs in Excel. Now, Copilot can act like a personal data describer or visual interpreter converting visual information into text. So let's say that you've received a quarterly sales Excel workbook and it contains multiple charts which have been created by your colleague. So here's how you could use Copilot to interpret those charts.
First, you would open the workbook in Excel and then press Alt + H + FX to open the Copilot pane. You want to commit that sequence to memory. The focus will go to the Copilot chat input field. If you need to move focus back to your spreadsheet after opening that Copilot pane, you can press F6 to cycle through the different options. Excel uses F6 to switch between the worksheet grid and also task panes like Copilot and the ribbon. Next, you want to ensure that you are on the worksheet that has the chart that you want to describe. So you can use control page up and control page down to move between sheets until you reach the correct worksheet.
When your focus lands on the chart or object, your screen reader, depending on which one you have, may just let you know that you're on an object. It may give you a generic description. You want to press F6 to jump back to Copilot. Then you're in the text box and you can type in a prompt, something like describe the chart on this worksheet including all data points and trends.
So you want to be quite precise about the prompt that you type in, and then you can press enter to send that to Copilot. You'll get a pause and Copilot will analyze the chart's underlying data and any labels and titles, and in a moment it will produce a detailed text summary and a description of the chart.
Copilot's answer will appear in its pane. For example, it might say something like, this worksheet contains a column chart titled Q3 sales by region. The chart shows five regions on the Y axis, north, south, east, west, and central. The Y axis represents sales in millions of dollars from zero to 50 million dollars. The north region has the highest sales at 45 million dollars, followed by east at 38 million, central at 28 million, west at 22 million and south at 18 million. There's an upward trend line indicating 15% growth quarter over quarter.
That level of detail gives you a mental picture of the chart's content. You might prefer examining exact numbers in a table, and you can ask Copilot, "Convert this chart into a data table I can navigate with my screen reader" and Copilot will then output the data behind the chart in a structured table form. For example, a table with columns and regions, different headings like North 45, et cetera, and just take you through a different data that you'll be able to navigate using table keystrokes. And that essentially gives you an on-the-fly accessible table version of the chart that was once not accessible.
You can ask for a summary of insights, so you might also want a high level summary instead of all the raw numbers. You could prompt Copilot, "Create a bulleted summary of the key insights from this chart," and if you issued a prompt like that, Copilot could list the main points. For example, that the north region is the top performer at 45 million in Q3 sales and significantly above other regions.
It could tell you the south region has the lowest sales at 18 million dollars, which is something that if you were the responsible person you'd also want to know about. You'd want to give the south region some tender loving care to get their numbers up. So you can then copy these points to your notes or include them in an email. Effectively, Copilot has helped generate a written analysis of the visual chart.
Now, let's turn to Microsoft Word and how Copilot can help with images, diagrams, or other graphics that you might find in a Word document. Often, coworkers will include charts and organizational diagrams and screenshots in a Word report without providing alt text. So if you find it inaccessible document and the organization you're working for has a policy on this, you should absolutely push back on that and say, "You should be running the accessibility checker. You shouldn't be giving me these inaccessible documents." But in the end, we got to get the job done.
So with Copilot, you can get those visual descriptions on demand when it's working. In Word, you can press Alt + H + FX once again to open the Copilot pane. That's the same as in Excel and many other places in Office. You can then press F6 to return to the document body. Navigate through the document as you normally would with your screen reader's reading keys usually up and down arrow or navigate by heading if it has them. When you reach an image or a graphic, you can ask Copilot to describe that image. So you would jump back to the Copilot pane using F6 or Shift FX to focus the Copilot pane and then type a prompt.
It could be something like describe the image where my cursor is currently positioned, including alt text, colors and visual elements, and then press Enter. It'll go away and think, and Copilot will analyze the image at the cursor location and in a few minutes, it will generate a detailed description for you. If the graphic contains embedded text like a screenshot or a slide or a scanned document, you can ask Copilot to OCR it, perform optical character recognition for you. For example, extract alt text from this image and give it to me as plain text. Copilot will use its vision capabilities to read any text that is in the image and output it.
It is extremely useful for images of charts with labels, with photos, with documents from whiteboards or any situation where important text wasn't accessible. You will usually get a text block in the Copilot pane that you can then review. It'll be in the virtual cursor or browse mode. You can copy that if you need to.
While we're in Word, let's also cover a very handy keystroke for drafting text with AI. Microsoft has a draft with Copilot feature in Word, and it works similarly when you're composing emails in Microsoft Outlook if you have Word set as the editor and you invoke this simply by pressing Alt with I. And Alt + I opens a dialogue where you can type a description of what you want to write and Copilot will generate text for you. Essentially, this is a quick way to have Copilot write a passage of text in place based on your prompt.
Make sure that your cursor is on a new blank line. You don't want your cursor in the middle of existing text because the shortcut's not going to work there. You might use this to get a text drafter of a paragraph to have Copilot rephrase a sentence or to brainstorm ideas. In Outlook's email composer, Alt + I can help draft an email reply or refine your wording so it is sensitive to the email that you are replying to. While it's a powerful tool, it is wise to treat the AI's output as a draft.
Make sure to fact check and tweak it so it truly reflects what you want to say. And also that it reflects your style. There's nothing quite so demoralizing as receiving a reply from someone and realizing that they thought so little of you that they just got AI to generate an automatic one. So try and make it like you wrote it.
Now let's flip the Excel scenario. So instead of interpreting a chart, now you want to actually create a chart to share with sighted colleagues maybe who are more visual learners. For instance, a visual chart of financial data, and Copilot can help you generate and format charts through simple language commands.
So let's say that you would enter the data that you want to be in your chart in an organized table format. Structure is really critical here. For example, let's say that you have monthly revenue and expenses to plot in an Excel sheet. Put clear headers in the first row. For example, cell A1 could be month, B1 could be revenue, C1 expenses, and then list the month's names in A2 through A7, revenue figures in B2 through B7 and expense figures in C2 through C7. Make sure that the data range is contiguous.
Then using the keyboard navigation, get to the start of your data, which would be A1 in this case, and you shift with your arrow keys or whatever other tools you have to select the whole table of data, headers and all rows. You want Excel to know where the cells start and stop. With that range still selected, now you go back to our friend, the Copilot pane with Alt + H + FX, and in the prompt box, type your request to create your chart, including any formatting preferences that you might like. You could say, "Create a professional column chart for my selected data.
Use a blue color for revenue columns and a green color for expenses columns. Add a title, Financial Performance Q3 2025 and include a trend line for the revenue" and then press Enter. Copilot will insert a chart object into the Excel sheet for you. Then you want, of course, to verify the output. So you can go back and do it in reverse. You can say, "Describe the chart that you just created, including the visual elements and how the data is displayed." And Copilot will give you a detailed description. So it works both ways.
Email can be another challenging area, especially when you have a long thread with many replies and participants. So let's discuss how Copilot can help summarize and extract key points from email in Microsoft Outlook. Navigate to an email conversation, for example, an Outlook messages list. You could open the email which might contain all the previous replies collapsed.
Once you have the email open in the reading pane or in a separate window, you want to make sure that the focus is in the message. Then you can press Alt + H + FX, just as you would in Word or Excel. And this should open the copilot pane within the Outlook window.
If Outlook has the message in a separate window, Alt + H +FX works there. If you're using the preview pane, then you may need to ensure focus is in the message body itself. In the Copilot text box, type a prompt like summarize this entire email thread, highlight all action items, decisions made, and any outstanding questions. Organize the summary by topic and include who is responsible for each action item. And then you can press Enter. Copilot will digest the multi email thread, which could be dozens of messages and produce a coherent summary.
It might output something with headings or bullet points for each topic, making it easy for you to navigate the extensive result. You can ask follow-up questions as well. For example, who in this thread has outstanding questions directed at them or show me the list of tasks with due dates mentioned and Copilot will contextually answer based on the thread information. One of Copilot's most powerful capabilities is something Microsoft calls Business Chat, and that's the ability to pull information together from across your Microsoft 365 suite, documents, emails, Teams, chats, and meetings.
So let's walk through a scenario of using copilot as a cross app research assistant. So for this example, we'll say that you are preparing for a project review meeting tomorrow, and you want to quickly gather everything that's happened in the past two weeks related to that project without manually searching through email, Teams, and SharePoint because discussion about those projects could have happened in multiple Microsoft applications, which can make gathering the data time-consuming.
You could go to copilot.microsoft.com or you could open Copilot and Microsoft Teams if that's available there. To use the conversational AI in a broad scope, you've got to make sure that you have focus in the Copilot chat input box, and then you could type a prompt, something like, find all messages, files and meeting notes relating to, let's call it, Project CENA from the last two weeks across all channels I have access to, and then press Enter.
In Copilot, it might take it a while, but Copilot will perform a graph search across your data thanks to the semantic index and graph API that's integrated with Microsoft Copilot, and then give you the results. And now you can follow up with something like, well, based on those sources, what are the key issues I should be prepared to discuss in the review meeting? And Copilot will analyze the content for those files and give you an answer that will hopefully make you seem on top of it and together with a meeting.
So let's conclude with a few final accessibility thoughts. Using Copilot alongside a screen reader is generally quite a smooth experience, and we do work at the National Federation of the Blind with Microsoft quite extensively on this. They're constantly interested in feedback and making the experience very smooth. They believe in Copilot and they're investing a lot in it.
There are sometimes a few quirks and challenges though that you should be aware of. Sometimes after Copilot generates an answer, your screen reader might not automatically start reading it. The Copilot pane is a dynamic web-based area and screen readers might not always detect that new text has arrived immediately.
So to get around this, you can manually refresh your screen reader's virtual buffer. For instance, in JAWS, you would press the JAWS key with Escape, which is the refresh screen command. In NVDA, you would press NVDA with F5 to do the same, and that forces the screen reader to rescan the application's user interface, and that should announce the newly appeared Copilot text.
Also, make sure that you are running the latest version of your screen reader because compatibility is improving rapidly. As Microsoft refines, Copilot and screen reader developers update their code. There is a close collaboration between Microsoft and the screen reader developers, particularly on all of this AI stuff.
The Copilot chat pane is essentially a web view inside the Office app, and that means that you may need to use browse mode or the virtual cursor to navigate its content just as you would on a web page. In JAWS, you want to ensure that the virtual cursor is on and in NVDA, you want to make sure that browse mode is enabled if it's not automatically toggled on.
Once in the right mode, you can use standard reading keys, your up and down arrow usually, and potentially hitting navigation commands, et cetera, to get around this structure. And through sections of Copilot, you will find that you can navigate quite efficiently if you just use commands like heading, navigate to the next form field, et cetera.
So that's an overview of what you can do with Microsoft Copilot. There is plenty of training material available to you as well.
When we come back, Kennedy Zimnik is going to talk with us about AI in Microsoft PowerPoint as Access On continues.
Speaker 7:
Check out my podcast, The People Make the Places. Can you guess who this famous actor is who passed away in 2025?
Speaker 8:
Imagine Mike May blinded at just three years old. Mike's lifelong attitude was one of navigating around or crashing through every obstacle. That curiosity and willingness to crash through carried Mike all over the world, meeting presidents and celebrities and starting his own company. He tried virtually every sport, but for Mike, skiing was something special.
Speaker 7:
Who is this famous artist?
Speaker 9:
And as I was listening to the many things said about OJ, I thought about Lula May because my mother was out of control.
Speaker 7:
Me in a village in Ghana in 1976.
I just ate it in my second gutter. I told you about these trenches they have around here, about a couple feet deep. I just fell in one. A car came roaring by. They don't give pedestrians the right of way. They just beep their horn and just keep on roaring. You better get the heck out of the way, or you've had it.
For more highlights, check out episodes 13 and 14 of the people make the places.
Jonathan Mosen:
Now let's go back to Kennedy, who's going to talk to us about PowerPoint.
Kennedy Zimnik:
Great, thanks so much, Jonathan. PowerPoint, traditionally for blind or low-vision users, has brought up a lot of feelings of fear. It's a very visual-based medium. Navigating PowerPoint wasn't always the easiest to do with screen readers. Adding images, how do I know where to place the image? How do I create alt text for the image if I can't see it? Copilot has brought a solution to all of these issues and gives a great starting point for anybody who wants to create a PowerPoint right within PowerPoint using Copilot.
So before I start sharing and show you a little bit of a demo, I'm going to show you how it can build a complete PowerPoint deck from either a simple prompt or an existing document. That's also another cool part. So as Jonathan's been describing, if you put in a prompt, then it can create something from that prompt.
In PowerPoint, you can also have it create something from a document. You can do that in Excel and Word as well. But PowerPoint is really useful because you can have a Word document with separate headings and then just say, "Use each separate heading as a different slide in this PowerPoint." That's really simple way to get 10 or 15 slides right off the bat, and then you can start editing.
So you can think of it as a fast first draft maker that you can reshape in minutes instead of hours. So like I said in the past, you used to have to go through and create each slide step by step and then go back and start editing. This saves you a ton of time and gives you a great jumping off point.
So first, I'll explain the workflow at a high level, and then I'll do two quick PowerPoints. First, from a blank prompt and then from a real document that I've created and used in past boutiques. It's the quick tour of Baltimore, Maryland. If you ever come to any of our document accessibility boutiques, you know what I'm talking about.
And then I'll go through and show you how to edit them and some guardrails for responsible use because just like with any AI, we don't want to completely rely on the information it provides. We want to verify everything that it produces. But like I said, it's just a starting off point. You wouldn't say produce a 20-slide PowerPoint and then not even look at it and then send it off to your boss or what have you.
So Copilot is great for starting from zero, repackaging or summarizing long documents into slides and creating a strong starting point. You should always double check the work and information Copilot or any other AI tool comes up with.
So before I start sharing my screen, I'll describe what I'm doing. I'll turn on a screen reader so you can hear what's going on. But when I don't have the screen reader on, I'll be very descriptive so you know what's going on on the screen. In PowerPoint, when you first open PowerPoint, the Copilot button is now the first button in the row of new. So you have a new heading and then you have create with Copilot, and then you have blank template, and then you have all the other templates that are normally there.
This is where you can pick from pre-made templates as starting points. I'm going to give a plain English prompt and let it draft the deck. So here's my prompt. Create a ten-slide presentation that explains what Microsoft Copilot is, who it's for, the key benefits, risks to watch, and a simple three-step adoption plan. Keep slides concise. Add speaker notes with examples. Suggest a clean modern layout. Also, add logical and concise alternative text to images and make sure slide reading order is logical.
So I know that's a mouthful, but the more specific and concise you are with the prompt originally, the better outcome you'll get at first.
Speaker 11:
Create with Copilot Alt + F + HY, 31 of 7.
Kennedy Zimnik:
So this was just the first option in this new, so it's create with Copilot. I'm going to press Enter.
Speaker 11:
Spell check, dialogue, Copilot narrative builder document, edit multi-line, create a presentation about ... describe the presentation you'd like to create and reference up to five files if needed.
Kennedy Zimnik:
I already have my prompt copied to my clipboard. I'm just going to paste that. But this is where you type what you want your PowerPoint to have, and this is also where you can include files, which I'll do in a second. For now, I'm just going to go off the prompt, okay? So I copy and paste to the prompt. I won't reread it. I'm asking to create a presentation that explains what Microsoft Copilot is.
Speaker 11:
Logical and ... reference file, send button, generating presentation topics, stop generating, ESC, adding more details, stop generating, ESC, main landmark, stop generating, ESC button, heading level three, main landmark, Microsoft Copilot, narrative topics list.
Kennedy Zimnik:
Once it's done generating, you get this high level overview of this slide that it created. You could arrow down and hear what it came up with.
Speaker 11:
Introduction to Microsoft Copilot, button ... list with two items, what is Microsoft Copilot, overview of its capabilities, out of ... button ... Microsoft button reorder topic two, target ... target audience for Microsoft.
Kennedy Zimnik:
There's also buttons that allow you to reorder topics or delete topics.
Speaker 11:
Button delete topic, list with three items, business professionals, developers and technical.
Kennedy Zimnik:
And we won't go through all the slides here for time, but a quick overview, each slide is target audience for Microsoft Copilot, key benefits, risks, and consideration. Three-step adoption plan. So everything I had in my prompt now shows up as slides.
Speaker 11:
Add more with ... reorder, delete, add more, delete add, generate slides button.
Kennedy Zimnik:
I just skipped down to the generate slides button. So now once you're happy with this starting point, with this high level overview, you can do generate slides.
Speaker 11:
Main landmark, stop generating, ESC button, heading level three, crafting a compelling narrative, stop generating, ESC, adding titles to slides, stop generating, ESC.
Kennedy Zimnik:
And you can hear it describe what it's doing as it's doing it. Now on the screen ...
Speaker 11:
Generating content for each slide, stop generating, ESC, generating content for each slide, pulling your slides together, creating your presentation, applying a design, presentation 1, PowerPoint, slide sorter view, Copilot grouping, Copilot grouping, keep it button.
Kennedy Zimnik:
As that was generating, a window popped up and it showed information populating itself. Now the focus is jump to a keep it button. So if you wanted to double check the slides before you hit keep it and then the regular PowerPoint pane view shows up, you could edit it here. For time, I'm just going to keep it because I'm happy with the slides that it created.
Speaker 11:
Presentation 1, PowerPoint, slide sorter view, slide 1.
Kennedy Zimnik:
Okay, so now I can go through and edit specific slides.
Speaker 11:
Slide 1, slide 3, introduction to Microsoft Copilot. Slide 2, meeting program. Slide 3, introduction to Microsoft Copilot.
Kennedy Zimnik:
Let's go to slide 3.
Speaker 11:
Copilot window pane, add in.
Kennedy Zimnik:
So there is a Copilot pane, just like Jonathan was describing in Excel, that pops up. So from what I could find, unlike Excel and Word, Copilot can't directly edit the PowerPoint that it created. Now, if I'm wrong, somebody please correct me, but I've been trying to get it to edit it, and it tells me each time it can't directly edit it. So this chatbot is here to help you, and you can just copy and paste things from the chatbot into your slide.
There is a new slide with Copilot button that you can choose. Like we were saying before, it's always smart to go through and double check the information in the slides. It might make things up or create wrong information, things like that. So this is where you would do that. You can check the alternative text of the images. You always want to check the alt text that it generates because sometimes it doesn't do the best job, especially when it comes to context.
So I'm going to turn off the screen reader just for a second.
Speaker 11:
Exit NV.
Kennedy Zimnik:
If I wanted to go through and edit these slides, I would do that, normally, switching panes using F6, using the arrow keys and then tab keys to get into each object. Something else that I wanted to talk about is using the accessibility checker. So if you were to go to file, I'm just using a mouse for right now, but if you're using a screen reader, you could do alt and then F to get to file and then scroll down and get to info, and then check for issues and then check accessibility. So I'm going to hit check accessibility.
Now, an accessibility assistant pane pops up on the right side. There are some issues. There are headings that list each issue and if there is an issue or not. So color and contrast, there's a check mark, meaning there are no issues. Media and illustrations, there is missing alt text for images so we want to go in and double check that. I won't do that right now.
If we wanted to go through and check the alternative text for each image all at one time, you go to the image, you right click on the image or do shift F10 to simulate a right click. Sorry. One of the last options is view alt text. Now you can go through each image and check that the alt text makes sense and is logical. So this specific image is digital work of smart mobile phone with social media followers. That might be an okay description.
To me, the image is showing somebody with many social media followers. It's a smartphone on a table. So I might switch that around. But this just goes to show you that sometimes alt texts can be okay if it's automatically created, but most of the time, you do need human interaction with that alt text in order to change it to something more logical.
I'm going to close this PowerPoint and not save it. And then I'm going to reopen PowerPoint and reshare my screen. So now I'm going to create a PowerPoint using a document that's already been created. So I'm going to turn my screen reader back on.
Speaker 11:
Search toggle, new PowerPoint, file list, home, new alt F, new grouping, new grouping, new grouping, one of one, list one of two, create with Copilot, alt F, and L. Slide 1, spell check, dialogue, Copilot narrative builder document.
Kennedy Zimnik:
Okay, so for the prompt, I'm going to say use the attached file to create a PowerPoint on a tour of Baltimore. Each heading should be its own slide. Then there's a reference files button that I'm going to navigate to using tab.
Speaker 11:
Reference files button, edit multi-line, create a presentation about ... use the attached file to create a PowerPoint on a tour of Baltimore. Each heading should be its own slide/type to reference a file and describe the presentation you'd like to create.
Kennedy Zimnik:
So this is an interesting way to do it. Instead of opening up a file manager, you just start typing whatever the file name is and it automatically pops up. So I'm going to type in a quick tour of Baltimore.
Speaker 11:
A Q-U-I-C-K, a quick tour of Baltimore doc accessibility demo.
Kennedy Zimnik:
So what's cool about this is it can do Word documents, which makes sense, but it can also do PowerPoint documents if it's made excessively. So I'm going to attach this PowerPoint document of a quick tour of Baltimore, Maryland.
Speaker 11:
Edit multi-line, create a presentation about a quick tour of ... attachments toolbar, reference files, send button.
Kennedy Zimnik:
I'm going to hit send.
Speaker 11:
Generating presentation topics, stop generating, ESC, main landmark, stop generating, ESC button, heading level three, gathering more ideas, stop ... main landmark.
Kennedy Zimnik:
Great. So again, I can use the arrow key to see this high level overview of the slides that it created. I'm just going to skip down for time.
Speaker 11:
Narrative topics, one button, delete ... add, reorder topic, one, delete, add more with ... reorder one, delete. Add more with ... reorder one, delete. Add more with Copilot, generate slides button.
Kennedy Zimnik:
To generate slides ...
Speaker 11:
Main landmark, stop generating, ESC button, heading level three, crafting a compelling narrative, stop generating, ESC, adding titles to slides, stop generating, ESC.
Kennedy Zimnik:
Now, it showed me the same screen that it did before and I'm going to hit the keep it button.
Speaker 11:
Speech mode talk, presentation one, PowerPoint.
Kennedy Zimnik:
Okay, now if I want to start editing ...
Speaker 11:
Slide 3, overview of Baltimore.
Kennedy Zimnik:
... I'm just going to press the arrow key and press Enter.
Speaker 11:
Copilot window pane.
Kennedy Zimnik:
I'm going to close this Copilot window.
Speaker 11:
Enter ... work toggle, web ... open close button. Slide 3, overview of Baltimore.
Kennedy Zimnik:
And there we go. I have all my different slides, so if I arrow down I can hear what they are.
Speaker 11:
Slide 4, introduction to Baltimore and its attractions. Slide 5, places to visit in ... Slide 6, national Federation of the Blind Headquarters, highlights and facilities. Slide 7, Baltimore Aquarium. Slide 8, dining in Baltimore. Slide 9, Phillips Seafood.
Kennedy Zimnik:
So it's also taking images from my PDF, which I think is really cool. I want to go in and double check the alt text.
Speaker 11:
Exit.
Kennedy Zimnik:
It looked like it screen grabbed the images from the PDF and combined them. So for images and specifically PDFs, it might not be the best at grabbing each individual object or element, like two images next to each other. They shouldn't be one image. Just be careful of that. That's just another way that AI assumes things are together when they're actually not. And then, of course, you could also run this one through the accessibility assistant and see if your PowerPoint is accessible.
Okay, that is going to do it for Copilot and creating PowerPoint presentations with Copilot. Jonathan, did you want to talk briefly about taking notes with AI and other cool AI things?
Jonathan Mosen:
I will, but first I've got to ask you, so does this mean that it's going to put a big pile of piping hot Maryland crab cakes in my OneDrive?
Kennedy Zimnik:
I wish. I wish.
Jonathan Mosen:
We've got to get that working. All right. I'm going to talk briefly about meetings because taking notes at meetings can be tricky when you're trying to concentrate on the meeting, participate in the meeting, and then you're supposed to take notes as well. And really, you don't have to do that anymore.
When you're doing online meetings, all of the big offerings in the meeting space, Zoom, Google Meet, and Microsoft Teams, have similar offerings, which is that you can enable AI to monitor the meeting for you. It can produce summaries and key action points, which you as the meeting organizer can have, or you can send them out to all meeting participants. They vary a bit, but the functionality is largely the same.
You can also have the AI enabled in a way that allows somebody who's come in late to the meeting to type quietly to the AI and say, "What did I miss? Catch me up?" or, "What did we just decide again?" and it can help to provide information.
So there's plenty of documentation online about those online AI offerings for those sorts of meetings. But of course, not all meetings happen online. AI can still help here either through specialized devices or using your phone to capture the audio, to transcribe it or even your smartwatch. So let's briefly look at some of these options.
Plaud is a new line of AI-powered voice recording devices and that is spelt P-L-A-U-D. And these are designed specifically to capture meetings and summarize them. The two notable products at the moment from Plaud are their Note Pro and their NotePin. The Plaud Note Pro is a credit-sized card device and it attaches to your phone.
It has four microphones and intelligent transcription capabilities. It automatically detects whether you're in a conference room or on a call and it adjusts accordingly. For example, it uses AI beamforming microphones to capture clear audio up to five meters away in a meeting room, and then it switches to a different mode for phone calls.
With the tap of a button, you can highlight important moments in real time and that tells the AI to pay special attention to those parts when it's summarizing later. After the meeting, the device generates a transcript and an action list summary focusing on decisions and next steps. So it's like coming out of Zoom or Teams or Gemini meeting, but it's an in-person meeting.
It supports 112 languages and it can label speakers, producing role-specific summaries, for example, tasks for marketing versus engineering, and you can export that to various formats or team tools. It has a long battery life, 30 to 50 hours recording time, and it stores audio locally on this credit card-sized device for later with cloud synchronization possible.
Now, Plaud NotePin is an even smaller wearable recorder and you can hear an evaluation and demonstration of Plaud NotePin all the way back on Access On episode 7. It's a USB stick-sized device. You can press it once to start recording and press it again and it will stop recording. It gives a vibration buzz feedback. One buzz when it starts and a double buzz when it stops. You can clip it to your shirt or you can wear it on a lanyard and you can discreetly record meetings or lectures hands-free.
After recording, again, you use the companion Plaud app on your iPhone or your Android device to access the audio, which is then transcribed and summarized by AI in the app. Now the accessibility of the Plaud app, as you will hear if you check out Access On episode 7, is not as good as it could be. But with a bit of persistence, if you're using things like screen recognition on your iPhone, you can make it work and you may determine that these devices are just so handy that it's worth the hassle.
Unfortunately, the National Federation of the Blind and several blind individuals have found Plaud responsive to feedback, but they haven't actually come through with substantial improvements to the accessibility shortcomings that we've identified at this point. But you don't need specialized hardware to use AI for note-taking. Your smartphone or even your smartwatch can serve as the recorder with cloud AI services doing the transcription.
What I've taken to doing, having become a bit disillusioned with the accessibility of the Plaud app, is I use my Apple Watch Ultra. I'm very pleased to be back in the world of Apple Watch Ultra again, and it has an action button. So I've assigned the action button to turn on the voice memos app and start recording. I then take that recording and the Apple Watch makes surprisingly good ones and send that to Google Gemini because Google Gemini can now take audio files and summarize or transcribe them, so that's very effective.
You might also want to try Otter AI for a similar thing, although there are some accessibility challenges with aspects of Otter AI as well. There are various other offerings in the market and what is little known is that Microsoft Word online has a feature that allows you to transcribe audio files from within Word online and have them sent straight to a Word document.
So that's actually quite a good tool, by the way, if you're trying to find something that will transcribe a podcast for you. It does a surprisingly good job in Microsoft Word online.
So that's a brief overview of ways that you can transcribe meetings, be they online or not, and I certainly find it very helpful. I went to the M‑Enabling Summit recently. I wanted to take notes, but I wanted to focus on the event rather than sit there typing away. And I was able to use my Apple Watch Ultra, send the presentations to Google Gemini and have really good summaries of the events that I attended. So when it all works, we live in a wondrous age.
And that concludes highlights of the webinar that we put together late last year on AI and employment. But just a quick postscript, I am pleased to report that while there are still some things to do, the Plaud apps accessibility has improved quite a bit, at least in iOS. And I did invest in a Plaud Note Pro. I like this device a lot. There is still work that I would like to see done, but it's better than it was. The battery life goes on for a long time and once it understands who speakers are, it just goes ahead and labels those speakers.
I used it to take notes recently for a strategy meeting that we had here at our Center of Excellence in Nonvisual Accessibility. We did about five or six hours of meeting that day and it produced amazing summaries and action points. Everybody was labeled. It was just fantastic what it did.
So it is one of those things where sometimes I wish still that it was better than it was, but it makes such a difference that I hang in there and it is good to see some marginal incremental improvement going on. If you are using a Plaud Note Pro or a Plaud NotePin, I'd be interested to hear how you're getting on with it and whether you find it worthwhile.
That concludes this episode of Access On, the technology podcast of the National Federation of the Blind. To send in a contribution for a future episode, email us. Attach an audio clip or just write it down and send it to [email protected]. That's [email protected]. To keep up to date with Access On, follow us on Mastodon, [email protected]. That's [email protected] on Mastodon.
To subscribe to an announcement-only email list about upcoming episodes, send a blank message to [email protected]. That's [email protected]. To learn more about the National Federation of the Blind, visit our website, nfb.org or phone us (410) 659-9314. That's (410) 659-9314. And be sure to check out the Nation's Blind Podcast right from where you heard this podcast.